NWRA Redmond, WA Associates, Inc. NorthWest Research Development of a New Probabilistic Wake Vortex Prediction Model Matthew J. Pruis NorthWest Research Associates WakeNet3-Europe Specific Workshop on “Operational Wake Vortex Models” Institute of Mechanics, Materials and Civil Engineering (iMMC), Division TFL, Université catholique de Louvain (UCL), 1348 Louvain-la-Neuve, Belgium 7-8 November 2011
33
Embed
Development of a New Probabilistic Wake Vortex Prediction ...€¦ · Wake Research And Prediction System (WRAPS) —Ensemble of deterministic, fast-time wake vortex prediction models
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
NWRARedmond, WA
Associates, Inc.NorthWest Research
Development of a New Probabilistic
Wake Vortex Prediction Model
Matthew J. Pruis NorthWest Research Associates
WakeNet3-Europe Specific Workshop on“Operational Wake Vortex Models”
Institute of Mechanics, Materials and Civil Engineering (iMMC), Division TFL,Université catholique de Louvain (UCL), 1348 Louvain-la-Neuve, Belgium
7-8 November 2011
Acknowledgments
— This work was sponsored by the National Aeronautics and Space
Administration Air Space Systems Program
— The work was performed under the NASA NRA “Enabling Super-Dense
Operations by Advancing the State of the Art of Fast-Time Wake Vortex
Modeling”
— The lidar data shown in this study was provided by NASA
— The Federal Aviation Administration (FAA) has also provided lidar data to
use in comparison studies and also provided funding for several of the
models that have been incorporated in the new probabilistic model
— Current funding for this work is under the NASA NRA “Wake Vortex Data
Collection for Robust Modeling Validation to Enable Advanced, NextGen,
Wake-Conscious, Capacity-Enhancing Concepts”
— Neil O’Connor and Dr. Fred Proctor are the technical monitors
2
3
What did we do?
and
What did we find out?
Next Steps
Motivation
Discussion on related scientific
questions, and the feasibility
and priorities of different
methods with respect to future
research
Goal of NASA Probabilistic Model
Quantify the probability of finding a vortex at a specified location at a specified time after
passage of a known aircraft in a known atmosphere, where the vortex is of a
specified strength or greater
4
Why is a Fast-time Probabilistic
Model Needed?
� Large number of model inputs
� All model inputs have uncertainties
� Interaction of these uncertainties and affect of
uncertainties on the model predictions is not obvious
� Not clear there is a “best” deterministic fast-time wake vortex prediction model
� Uncertainties in observations are poorly quantified
5
6
Suboptimum weather informationa) Uncertainties in meteorological
sensing
b) Nonhomogeneity of weather conditions
c) Changes in weather since time of last
observation
7
Uncertainties in aircraft parameters,
e.g., position
Also,
o aircraft weight
o true air speed
o initial vortex spacing
o time of overflight
2000
3000
4000
5000
6000
7000
8000
9000
-1000100
100200300400500
95% confidence intervals at 2.2 nm from threshold
Approximate dimensions of B752 shown for scale
Shown is the 95% confidence intervals estimated from the maximum standard deviations for each distance from the threshold reported in
[1] Zhang Y, Shortle J, Sherry L., 2010. Comparison of Arrival Tracks at Different Airports. In: Proceedings of 4th International Conference on Research in Air Transportation. Budapest, Hungary.
[2] Hall, T., M. Soares. 2008. Analysis of localizer and glide slope flight technical error. 27th Digital Avionics Systems Conference, St. Paul, MN.
8
Ambiguities due to model assumptions,
simplifications, and parameterizations
Environmental
ConditionsIn-Ground Effect
N* = 0.25
N* = 0.5
N* = 1
N* = 0
Out-of-Ground Effect
EDR* = 0.1
� Mean biases of the wake observations with LCMT pulsed lidar are
� 2-4 m in vertical
� 4-8 m in horizontal
9
Uncertainties in the observation of wakes
Lai and Delisi, 2010. Assessment of Pulsed Lidar Measurements of Aircraft Wake Vortex Positions Using a Lidar
Simulator, AIAA-2010-7988, AIAA Atmospheric and Space Environments Conference, Toronto, Ontario, Aug. 2-5.
10
Environmental Parameters• air density(z)• crosswind(z)• headwind(z)• EDR(z)• potential temperature(z)
Aircraft Parameters:• aircraft weight • air speed• initial vortex spacing• initial lateral position• initial vertical position
Used for Model Assessmentand
Calibration (Model Improvement)
Observations
Estimates of Wake Vortex Transport and
Decay
Fast-time Wake Vortex Prediction Models
Deterministic Fast-time Wake
Vortex Prediction Model
Wake Research And Prediction
System (WRAPS)
— Ensemble of deterministic, fast-time wake
vortex prediction models
— Input uncertainties (i.e., a/c position and speed,
weather) are included using a Monte-Carlo
approach
— Allows user to compare with wake data that
was obtained under similar conditions
11
Pruis and Delisi, 2011. Comparison of Ensemble Predictions of a New Probabilistic Fast-Time Wake Vortex Model
and Lidar Observed Vortex Circulation Intensities and Trajectories, AIAA-2011-3036, 3rd AIAA Atmospheric Space
Environments Conference, Honolulu, Hawaii, June 27-30.
How is this model different?
� Principal difference between the new NASA probabilistic model and other probabilistic models is that it is an ensemble of many deterministic fast-time wake vortex prediction models
12
13
Run deterministic fast-
time wake vortex
prediction model
Access and process
field data
Make
suggestions
1 more Monte Carlo simulation
Model set-up
Display and
Save Results
Model output data
files (optional
write for
documentation)
Model input data
files (optional
write for
documentation)
Set-up input for each
simulation
User request to compare with data
1 more deterministic model
Graphical User Interface
qc check reveals inconsistency with inputs
Wake Research
And Prediction
System
(WRAPS)
Fast-time models1. APA v3.22. APA v3.33. APA v3.44. D2P v1.0 (NWRA)
Air density at initial height of vortex 1.02 kg/m3 0.02
crosswind -0.6 m/s 2.6
headwind 0 m/s 0.44
EDR 1×10-5 m2/s3 Factor of 2
Potential temperature 300 °K 1
Potential Temperature Gradient 0 °C/m 0.0025
Aircraft weight 76340 kg 4400
Approach speed 64.8 m/s 3.6
Initial vortex spacing 29.8 m 1
Initial lateral position 0 m 9
Initial vertical position 217 m 9
17
WRAPS Simulation With One Model (APA v3.2)
18
Individual simulations
Gray region contains 90% of simulations
Red region contains
50% of simulations
Explanation of Simulation Results With One Model (APA v3.2)
19
WRAPS Simulation With One Model (APA v3.2)
20
WRAPS Simulation With Two Models (APA v3.2 and APA v3.4)
21
WRAPS Simulation With Three Models (APA v3.2, APA v3.4, TDP v2.1)
22
WRAPS Simulation With Four Models (APA v3.2, APA v3.4, TDP v2.1, and VPR v2.0)
Adding Additional Models to Ensemble Tends to Increase Model Spread
Slightly and Modifies the Ensemble Mean Only Slightly
23
A comparable data set of LMCT lidar observations
of wake vortex circulation intensity and trajectories
is 87 B752 landings at Denver airport in 2003
(TLast > 80 seconds)
Comparison with data
Should correspond to low turbulence conditions
24
87 landings of a B752 at DEN 2003
25
Overlay of the Model Predictions and Similar Wake Observations
What did we find out?
» Spread between different deterministic models overlaps when small, plausible uncertainties in aircraft and environmental conditions are used
» Model reproduces observed vortex behavior and predicts approximately the same mean and spread as the observations
26
Next Step(s)� Collect more high quality wake
observations, with good weather
and aircraft observations
27
� Better estimates of
uncertainties (or expected
variability) in aircraft and
weather inputs
28
Next Step(s)
How much do aircraft weights vary?
How do the weights vary, is it
predictable?
How much variance is there on the
aircraft position relative to the glide
slope path? How does it vary?
Can we use ADS-B to get a better
estimate?
What is the relevant timescale to
estimate turbulence for aircraft
wakes?
What are the errors associated with
using weather observations
obtained in different locations then
where the wake is observed and
modeled, and observations
obtained at different times or
averaged over the wake lifetime?
What are the errors in weather
sensing? How do we estimate
this?
� Model Improvements
o Wake observation lifetimes
o High stratification
o Low turbulence, weak stable stratification
29
Next Step(s)
Descent of Wakes modified by
stratification
Pruis and Delisi, 2011. Assessment of Fast-Time Wake
Vortex Prediction Models using Pulsed and Continuous
Wave Lidar Observations at Several Different Airports,
AIAA-2011-3035, 3rd AIAA Atmospheric Space
Environments Conference, Honolulu, Hawaii, June 27-30.
Pruis and Delisi, 2011. Correlation of the Temporal
Variability in the Crosswind and the Observation Lifetime of
Vortices Measured with a Pulsed Lidar, AIAA-2011-3199,
3rd AIAA Atmospheric Space Environments Conference,
Honolulu, Hawaii, June 27-30.
Observation Lifetime of Wakes, by EDR*
NASA has Funded a New Project
� Lack of data was impetus for a new NASA NRA entitled “Wake Vortex Data Collection for Robust Modeling Validation to Enable Advanced, NextGen, Wake-Conscious, Capacity-Enhancing Concepts”
30
Overview of New NASA NRA (1 of 2)
� Characterize existing sensors and sensing
capabilities
� Establish full set of wake vortex, meteorological,
aircraft, and air traffic operational parameters
required to be measured and test conditions for
several test scenarios
� Develop a ground-based, terminal-area, data
collection test program to collect data that can
be used to validate existing wake vortex
prediction tools31
� Collect meteorological and wake vortex position
and strength data
� Conduct a robust validation of wake vortex
models. Enhance, as required, existing fast-time
wake vortex prediction tools using the new data.
This includes both deterministic and
probabilistic tools that can:
� Predict the probability of wake location
� Predict the probability of wake location and strength